Skip to content

Compare models

Put up to 4 models beside each other — token prices, context windows, capabilities and provider, from the same catalogue the model pages read.

  1. Nemotron 3.5 Super VL (Free)NVIDIARemove
  2. Nemotron 3.5 Super VLNVIDIARemove
  3. KAT-Coder-Pro V1KwaipilotRemove
nemotron-3.5-super-vl:free vs nemotron-3.5-super-vl vs kat-coder-pro
AttributeNemotron 3.5 Super VL (Free)nemotron-3.5-super-vl:freeNemotron 3.5 Super VLnemotron-3.5-super-vlKAT-Coder-Pro V1kat-coder-pro
Pricing
Input$0 / 1M$0.10 / 1M$0.005 / 1M
Output$0 / 1M$0.40 / 1M$0.005 / 1M
Cache Write$0 / 1M——
Cache Read$0 / 1M$0.10 / 1M$0.005 / 1M
Cache Write (5m)—$0.10 / 1M$0.005 / 1M
Cache Write (1h)—$0.10 / 1M$0.005 / 1M
Web Search—$0 / 1M$0 / 1M
Context
Max context1M1M256K
Max outputN/AN/AN/A
Capabilities
VisionYesYesNo
Function CallingYesYesYes
JSON ModeYesYesYes
StreamingYesYesYes
Catalogue
ProviderNVIDIANVIDIAKwaipilot
Categorychatchatchat
Charge typeFreePay As You GoPay As You Go
Released———
Description
SummaryNVIDIA Nemotron 3.5 Super VL is an open multimodal Mixture-of-Experts (MoE) model featuring 120B total parameters with 12B activated per token. Built on a hybrid Mamba–Transformer architecture with multi-token prediction, it supports text, images, documents, GUI screenshots, and video inputs, along with a 1M-token context window for complex, long-horizon workflows. Optimized for multimodal reasoning, visual understanding, and agentic tool use, the model supports configurable reasoning modes and token budgets. It is well suited for document intelligence, web research, coding validation, customer service, GUI analysis, and video understanding, enabling efficient, long-running multimodal agent applications.NVIDIA Nemotron 3.5 Super VL is an open multimodal Mixture-of-Experts (MoE) model featuring 120B total parameters with 12B activated per token. Built on a hybrid Mamba–Transformer architecture with multi-token prediction, it supports text, images, documents, GUI screenshots, and video inputs, along with a 1M-token context window for complex, long-horizon workflows. Optimized for multimodal reasoning, visual understanding, and agentic tool use, the model supports configurable reasoning modes and token budgets. It is well suited for document intelligence, web research, coding validation, customer service, GUI analysis, and video understanding, enabling efficient, long-running multimodal agent applications.KAT-Coder-Pro V1 is KwaiKAT’s most advanced agentic coding model, built specifically for real-world software engineering. It achieves a 73.4% solve rate on SWE-Bench Verified and is optimized for tool use, multi-turn workflows, instruction following, and generalization through a multi-stage training pipeline that includes SFT, RFT, and large-scale agentic RL.